Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

AI/ML Engineer with 3+ years of experience building production machine learning and generative AI systems in financial services and enterprise data platforms. Experienced in designing LLM-powered applications, Retrieval-Augmented Generation (RAG) pipelines, transformer-based NLP, vector retrieval systems, and scalable ML infrastructure using Python, PyTorch, and TensorFlow. Skilled in LLM orchestration and retrieval frameworks (LangChain, LangGraph, LlamaIndex), hybrid retrieval and reranking, vector databases (FAISS, Pinecone, etc.), and MLOps for deployment, monitoring, and scalable inference using Docker, Kubernetes, and cloud services.…AI/ML Engineer with 3+ years of experience building production machine learning and generative AI systems in financial services and enterprise data platforms. Experienced in designing LLM-powered applications, Retrieval-Augmented Generation (RAG) pipelines, transformer-based NLP, vector retrieval systems, and scalable ML infrastructure using Python, PyTorch, and TensorFlow. Skilled in LLM orchestration and retrieval frameworks (LangChain, LangGraph, LlamaIndex), hybrid retrieval and reranking, vector databases (FAISS, Pinecone, etc.), and MLOps for deployment, monitoring, and scalable inference using Docker, Kubernetes, and cloud services.

Jhanavi Putcha

AI Engineer, Data Scientist, Data Engineer, +3





AI/ML Engineer with 3+ years of experience building production machine learning and generative AI systems in financial services and enterprise data platforms. Experienced in designing LLM-powered applications, Retrieval-Augmented Generation (RAG) pipelines, transformer-based NLP, vector retrieval systems, and scalable ML infrastructure using Python, PyTorch, and TensorFlow. Skilled in LLM orchestration and retrieval frameworks (LangChain, LangGraph, LlamaIndex), hybrid retrieval and reranking, vector databases (FAISS, Pinecone, etc.), and MLOps for deployment, monitoring, and scalable inference using Docker, Kubernetes, and cloud services.…AI/ML Engineer with 3+ years of experience building production machine learning and generative AI systems in financial services and enterprise data platforms. Experienced in designing LLM-powered applications, Retrieval-Augmented Generation (RAG) pipelines, transformer-based NLP, vector retrieval systems, and scalable ML infrastructure using Python, PyTorch, and TensorFlow. Skilled in LLM orchestration and retrieval frameworks (LangChain, LangGraph, LlamaIndex), hybrid retrieval and reranking, vector databases (FAISS, Pinecone, etc.), and MLOps for deployment, monitoring, and scalable inference using Docker, Kubernetes, and cloud services.

Available to hire

AI/ML Engineer with 3+ years of experience building production machine learning and generative AI systems in financial services and enterprise data platforms. Experienced in designing LLM-powered applications, Retrieval-Augmented Generation (RAG) pipelines, transformer-based NLP, vector retrieval systems, and scalable ML infrastructure using Python, PyTorch, and TensorFlow.

Skilled in LLM orchestration and retrieval frameworks (LangChain, LangGraph, LlamaIndex), hybrid retrieval and reranking, vector databases (FAISS, Pinecone, etc.), and MLOps for deployment, monitoring, and scalable inference using Docker, Kubernetes, and cloud services.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Intermediate

Work Experience

AI/ML Engineer at BNY New York, USA

May 1, 2025 - Present

Designed and deployed LLM-powered document intelligence pipelines using LangGraph, LlamaIndex, and GPT-4 to automate regulatory filings and compliance document processing, reducing manual review effort by 41%. Built a RAG platform integrating FAISS and Pinecone with financial data lakes, implementing hybrid vector search and reranking to enable low-latency retrieval for trade and settlement records; improved analyst efficiency by 29%. Fine-tuned transformer-based NLP models (BERT, FinBERT, LLaMA) for named entity recognition and document classification with 95% precision in financial entity extraction. Developed scalable data pipelines using PySpark, Apache Airflow, and Snowflake to process multi-terabyte transaction datasets, enabling near real-time analytics and 2.1× throughput. Trained transformer-based models for financial text classification and sentiment analysis with 26% accuracy improvement. Implemented anomaly detection using XGBoost, autoencoders, and graph-based analytics t

Machine Learning Engineer at LTIMindtree, Hyderabad, India

January 1, 2022 - July 31, 2024

Architected an AI-driven credit risk scoring system using XGBoost, LightGBM, and logistic regression trained on 12M+ loan records, improving default prediction accuracy from 74% to 91% and speeding underwriting decisions. Engineered a real-time fraud detection pipeline using gradient boosting, isolation forest, and graph-based anomaly detection to analyze card transaction streams, reducing fraud losses by 38% with false positives below 3%. Implemented MLOps workflows with Azure ML, MLflow, and Docker to automate training, versioning, and deployment, reducing release cycles from weeks to days while supporting regulatory compliance. Built LSTM and temporal CNN models to forecast account activity and detect suspicious account takeover behavior, improving early fraud detection by 27%. Established model monitoring dashboards in Power BI using Azure ML APIs to track drift and stability, enabling proactive retraining and reducing degradation by 30%. Developed transformer-based NLP pipelines (

Machine Learning Engineer at LTIMindtree

January 1, 2022 - July 1, 2024

Architected an AI-driven credit risk scoring system using XGBoost, LightGBM, and logistic regression on 12M+ loan records, improving default prediction accuracy from 74% to 91% and enabling faster underwriting decisions. Engineered real-time fraud detection using gradient boosting, isolation forest, and graph-based anomaly detection on card transaction streams, reducing fraud losses by 38% while keeping false positives below 3%. Implemented MLOps workflows using Azure Machine Learning, MLflow, and Docker to automate model training/versioning/deployment, reducing release cycles from weeks to days while ensuring regulatory compliance. Built deep learning models (LSTM and temporal CNN) for account activity forecasting and early fraud/account takeover detection, improving early detection by 27%. Established monitoring dashboards in Power BI integrated with Azure ML APIs to track data drift and performance, enabling proactive retraining and reducing degradation by 30%. Developed transformer